Research On Asteroid Surfae Image Segmentation Based On Swarm Intelligence Algorithm

As an important way of space technology innovation,deep space exploration has gradually become an important part of the future space program.In order to obtain more comprehensive and effective experimental data and information during the exploration mission,it needs to perform a soft landing in an unfamiliar terrain area with a complex environment.The autonomous navigation and environmental analysis of the lander are technical problems that need to be faced in the remote operation environment.Automatically detect and identify rocks in celestial bodies,so as to avoid obstacles,select special rocks for sampling and analysis,which can greatly improve the efficiency of celestial body exploration.Natural landforms such as gravel piles,basins,and craters of various sizes are often distributed on the surface of asteroids.The complex terrain environment will bring great difficulties to the exploration mission.Because the asteroid’s surface environment is covered by sand and dust for a long time,and it is affected by lighting and shooting conditions,it is difficult to distinguish the color of the rock mass from the background surface.In the problem of rock detection,the detector only relies on the target based on image processing on the earth’s ground.Detection method and real-time obstacle avoidance command may cause poor application effect on the asteroid surface or large communication delay,which leads to mission failure.Therefore,whether the lander can autonomously detect obstacles during soft landing is very important,and image segmentation It is the key issue of obstacle detection.In this paper,a series of researches on the segmentation technology,feature extraction and tracking of surface images based on the swarm intelligence hybrid algorithm are carried out.The main work content is as follows:(1)Research the traditional swarm intelligence algorithm,propose a hybrid optimization of particle swarm and gray wolf algorithm,and construct a fitness function based on two-dimensional information entropy of the image,and finally compare the algorithm in this paper with the traditional swarm intelligence algorithm through experimental simulation,and use Test function to verify the accuracy and stability of the algorithm.(2)Exploring the 3D reconstruction technology based on the asteroid model,and generating simulated surface images based on Open GL and the open source terrain model,and finally extracting and tracking the features of the image based on the optical flow method.(3)Apply the swarm intelligence hybrid algorithm mentioned above in image segmentation processing.The algorithm proposed in this paper can customize the number of segments of the segmentation threshold by changing the population initialization conditions,so as to meet the needs of multi-threshold segmentation in different scenes.Traditional segmentation algorithms can better predict flat terrain areas.